logIt » job automation

Nebri Automation for Weather Feed to Twitter

Arif — Tue, 02 Jun 2015 14:13:39 +0000

The season was intermittently rainy or dry as I began to take interest in weather. After getting stuck with “If-this-than-that” (IFTTT) for multiple location feeds of weather, I found out that I wasn’t alone, a geek hit the same wall, yet bringing another automation forgery: Nebri OS, an event-driven development platform based-on writing rules in Python. Without despise, eventually I still use IFTTT on the other end to send weather alert to my smartwatch again.

A quick adaptation of Nebri’s straightforward blog post is to first shortlist rain related codes into 15 of them and changing from forecast to latest measurement instead. Bridging Nebri and my smartwatch are Twitter and IFTTT recipe–the watch merely mirrored my phone, you don’t need it actually. Why (again) Twitter? Well, rather than the “then that” side–the watch enabler, the “if this” side of IFTTT for Twitter recipes is quite powerful given so many tweet filtering options.

Key-value pairs as trigger for event driven Nebri OS workflow

The workflow starts with Drips, a scheduler (cron) where key value-pairs (KVP) is created at certain time. A rule script (yweather) is triggered when this KVP is created. The weather results are sets of later KVPs feeding the twitter rule script where our simple Nebri workflow ends. This is how I get multiple cities, by using different Drips, different events, running the same workflow. Both yweather and twitter scripts are also example of API connectivity to Nebri.

Many things already taken care by Nebri, hence unseasoned programmer should easily cope with writing simple Python rule and focus on the automation, nevertheless complex rule should also be accommodated. I was having trouble on using previous KVP from different PID though–moved them to comment in the script, but they seem to be working on it. Check their YWeather blog post on how the basics work and compare what I did in Github. I embed an example tweet from @twithujan below. Yahoo! Weather frequency of measurement and accuracy are of different topic by the way.

Bogor is Light Rain with temperature at 23 °C. Taken at Thu, 28 May 2015 7:00 pm WIT

— Twit Hujan (@twithujan) May 28, 2015

PS: “hujan” means “rain” in Indonesian

Get Commuter Train Trouble Alert Delivered to Watch by IFTTT

Arif — Sat, 17 Jan 2015 14:31:02 +0000

Imagine getting alerts of Commuter Line train service disruption right on your wrist! Well, the watch thing isn’t a must as any SMS-enabled phone will do. But without the watch, I won’t be introduced to “If-this-than-that” (IFTTT), a simple logic line that glues popular apps into trigger and action branded as recipes. Anyone can code that line, a no-coding logic that cooks: sync wallpaper with Nat Geo’ instagram, volume goes up when Mom calls, change bulb light color after some space station moves, etc. As personal recipe, bring COOKOO connected watch to the table and I’ve leveraged public service disruption–no matter how ironic–into edgy personal business.

If this than that recipes alert for Commuter Train trouble

The ingredient is tweet, Twitter Search use case is no longer just sentiment analysis, its function extends to alert, even close to real time–I’m pretty optimistic given the number of Commuter Line users joining the conversation. Now, the no-code part of our recipe, a Twitter search query, that is

(to:krlmania OR @krlmania OR from:CommuterLine)
AND (gangguan OR pantograf OR wesel OR patah)
-semoga -smoga -moga

and a brief explanation is that it queries for some keywords for potential trouble within mentions/replies to @krlmania and official announcements of @CommuterLine. Exclusion added to the search, the sentiment part of tweets, expressing wishes to shoo trouble away–I’m sorry, but keep praying online anyway.

When Twitter launched Twitter Stories in 2011, I read about Ravi Pina running the crowd sourced @Caltrain. Later now I see that official accounts like @SMRT_Singapore is also a model of service disruption info center. We can go deeper to better stats from the API web service that shows real time position (an app is out there showing that the location data API is working), but until its open data status is known, crowd sourcing stays as the most reliable solution, independent to regulation, and easiest to implement. Does the recipe work? During the first week of testing, alerts were flooding–within limit of course–as trouble really happened. So, cook your own ingredients or get my recipes served here.

Train trouble alert on my watch

Queue Job To Be Executed At Certain Times

Arif — Mon, 30 Mar 2009 07:41:38 +0000

A series of command can be queued by using at command. There are slight variations between UNIXs for at command, the following example is for Ubuntu.

Due to to the purpose of running it for OpenView managed nodes (this is some sort of trial-test), I put the job queue as if it was done through ovdeploy (remotely).

$ echo "cat /etc/hosts > tst.txt" >> tst.at.queue

the queued job is a simple cat directed to an output file (tst.txt).

This next at will queue whatever contained in tst.at.queue

$ at -f tst.at.queue now + 2 minutes
warning: commands will be executed using /bin/sh
job 6 at Mon Mar 30 22:40:00 2009

We can check at queues as:

$ atq
6       Mon Mar 30 22:40:00 2009 a arif

It says that the job will be run at a certain time which is at 22:40:00 system time. When the job is done, queue will become empty and tst.txt will be generated as described in job file tst.at.queue.